Conditional Probability

Half of these 10 emails are spam. But once you know an email has a link, only the linked ones matter:

The share inside that group is the conditional probability of spam given a link, P(spam | link). Count both, then divide by the condition:

emails = pd.DataFrame({
    "link": ["yes", "no", "yes", "no", "no", "no", "yes", "no", "no", "yes"],
    "spam": ["yes", "no", "yes", "no", "no", "yes", "no", "no", "yes", "yes"],
})
spam = emails["spam"] == "yes"
link = emails["link"] == "yes"
print(spam.mean())
print((spam & link).sum() / link.sum())

This prints:

0.5
0.75

The condition you already know goes below the line: you divide by it. A spam filter works this way: the chance of spam given an email's words.